About

Why this exists

I've been working on several projects this year which require deep integrations with popular agent harnesses.

That led to a lot of fragmented research and documentation spread across different repos, which quickly became hard to manage.

In August, I decided to centralize it all in one repo. From that point, whenever I needed an answer to "can [harness] do [thing]", I would ask an agent in that repo.

The goal became "make something like MDN for harnesses" and you're now looking at it.

Slop disclaimer

This page was 100% written by a human.

Every other word on this site was written by GPT-6.1 Sol or Opus 5.5.

It is certain to contain crimes against the English language, UX faux pas, and blatant factual errors.

Despite this, it is likely very useful to a small group of people.